DP
RIETI Discussion Paper Series 10-E-048
Financial Distress and Industry Structure:
An inter-industry approach to the "Lost Decade" in Japan
OGAWA Kazuo
Institute of Social and Economic Research, Osaka University
Elmer STERKEN
University of Groningen
TOKUTSU Ichiro
Kobe University
The Research Institute of Economy, Trade and Industry
http://www.rieti.go.jp/en/
RIETI Discussion Paper Series
10-E-048
September 2010
Financial Distress and Industry Structure:
An Inter-industry Approach to the “Lost Decade” in Japan*
Kazuo Ogawa
Institute of Social and Economic Research, Osaka University
Elmer Sterken
Department of Economics, University of Groningen
Ichiro Tokutsu
Graduate School of Business Administration, Kobe University
Abstract
This paper proposes a novel approach to investigating the propagation mechanism of balance
sheet deterioration in financial institutions and firms, by extending the input-output analysis. First,
we use input-output tables classified by firm size. Second, we link the input-output table with the
balance sheet conditions of financial institutions and firms.
Based on Japanese input-output tables, we find that the lending attitude of financial institutions
affected firms’ input decision in the late 1990s and the early 2000s. Simulation exercises are
conducted to evaluate the effects of changes in the lending attitude toward small firms, as favorable
as toward large firms, on sectoral allocations. We find that output was increased for small firms and
reduced for large firms. The change in output was non-negligible, about 5.5% of the initial output of
each sector. In particular it exceeded 20% in textile, iron and steel and fabricated metal products.
Keywords: Input-output analysis, Trade credit, Balance sheet, Multiplier
JEL classification: C67, E23, E44, and L14
RIETI Discussion Papers Series aims at widely disseminating research results in the form of professional
papers, thereby stimulating lively discussion. The views expressed in the papers are solely those of the
author(s), and do not present those of the Research Institute of Economy, Trade and Industry.
*
This research was financially supported by Grant-in-Aid for Scientific Research (#19330044) from the
Japanese Ministry of Education, Culture, Sports, Science and Technology in 2009. We are grateful to
Masatoshi Yokohashi and Maki Tokoyama of Applied Research Institute Japan for kindly providing us
with the input-output tables classified by firm size. We are also grateful to Mary Amiti, Kosuke Aoki,
Jenny Corbett, Masahisa Fujita, Hermann Heinz, Hideaki Hirata, Yuzo Honda, Charles Horioka, Takeo
Hoshi, Takashi Kamihigashi, Anil Kashyap, Ryuzo Miyao, Arito Ono, Shigenori Shiratsuka, Nao Sudoh,
Wataru Takahashi, Kazusuke Tsujimura, Takayuki Tsuruga, Hiroshi Uchida, Iichiro Uesugi, Tsutomu
Watanabe, David Weinstein, Peng Xu and the participants of the 20th conference of Pan Pacific
Association of Input-Output Studies, Kobe Bubble and Financial Crisis Workshop,
NBER-CARF-CJEB-AJRC Japan Project Meeting and the seminars at Deutsch Bundesbank, Bank of
Japan, RIETI and Post-Keynesian meeting for extremely helpful comments and suggestions. Any
remaining errors are the sole responsibility of the authors.
1 Introduction
In Japan, the period of the 1990s and the early 2000s is called the ”Lost Decade,”
and in it the balance sheets of financial institutions and firms deteriorated greatly.
Many studies report that this had perverse effects on firms’ activities.1 This paper
investigates the effects of the balance sheet deterioration of financial institutions
and firms on the inter-industry structure. Input-output analysis is a powerful tool
for examining the inter-industry relationship from the general equilibrium viewpoint. Employing this input-output technique, this paper investigates how the balance sheet deterioration of financial institutions and firms are propagated across
sectors, and then evaluates quantitatively the extent to which the sectoral distribution is affected by balance sheet deterioration.2
Our study is related to the two strands of the literature. First, there is a growing
literature of multisectoral general equilibrium models that are intended to explain
the transmission of sectoral shocks through input-output linkages. This literature
includes Long and Plosser (1983), Basu (1995), Hornstein and Praschnik (1997),
Horvath (1998, 2000), Dupor (1999), Bergin and Feenstra (2000), Huang and Liu
(2001) and Shea (2002).
Secondly there are studies shedding light on the transmission mechanism of
sectoral shocks through credit chains. To illustrate in this framework how a deterioration in the balance sheet of one firm is transmitted to other firms through
inter-industry credit chains, suppose that customer A is hit by liquidity shock. The
supplier B will withhold completion of goods ordered from the customer A. Thus
1
For example, see Nishimura et al. (2005), Caballero el al. (2008), and Ogawa (2003a,b) for
the effects of balance sheet deterioration on firms ’entry and exit, and investment and employment,
decisions.
2
Kobayashi and Inaba (2002) analyze the propagation mechanism of coordination failure in one
sector triggered by non-performing loans in the banking sector, but this approach does not take full
advantage of input-output tables, whereas ours does. Tsuruta (2007) investigates whether credit
contagion leads to a decrease of trade credit supply for small businesses, using the micro data of the
Credit Risk Database. Tsuruta’s study does not take full interplays among sectors into consideration.
1
the supplier B will also run into liquidity problems, which in turn will affect the
suppliers that provide the supplier B with intermediate goods. In this manner, an
output reduction in one industry resulting from balance sheet deterioration may be
propagated into other industries, and thus eventually affect aggregate production.
Kiyotaki and Moore (1997, 2002) are pioneering studies, which show that a small,
temporary shock to the liquidity of some firms generates a large, persistent fall in
aggregate activity. Boissay (2006) and Raddatz(2008) are studies along this line.
The discussions above illustrate the importance of taking the inter-industry
linkages into consideration when investigating the propagation of financial distress
in one sector across sectors.3 Our study is on the same track with the two strands of
the literature in the sense that we investigate the propagation mechanism of balance
sheet shocks in one sector into the other sectors based on the input-output tables.
We extend the conventional input-output analysis in two directions. First, we
use input-output tables classified by firm size for the manufacturing sector.4 Specifically, the input structure of the j-th industry from the i-th industry is described by
four input-output coefficients, rather than one, as in the conventional input-output
table, because the input and output sectors are each divided into large and small
firms. Thus we obtain much richer information on the inter-industry relationship
than a conventional input-output table provides. The information in input-output
tables classified by firm size is very useful in analyzing the inter-industry structure
of the lost decade in Japan, because it is often argued that the balance sheet deterioration of financial institutions forced small firms to rely more on trade credit from
large firms, in order to meet their financial needs.
It is a tacit assumption underlying the credit chain argument that the firms
3
In a slightly different context, Lang and Stulz (1992) and Hertzel et al. (2008), using bankruptcy
filings data, examine the extent to which distress and bankruptcy filing have valuation consequences
for suppliers and customers of filing firms. However, they are silent on the macroeconomic consequence of financial distress.
4
Shimoda et al. (2005) is the only study that analyzes the industrial structure in Japan based on
input-output tables classified by firm size.
2
hit by liquidity shocks are credit-constrained. It is true that small-sized firms are
liquidity-constrained, but large firms have ample liquid assets to absorb the liquidity shocks coming from default of their customers. The upshot is that credit
contagion might be cushioned to some extent by the existence of large suppliers in
a network of firms. We can examine this possibility, using the input-output tables
classified by firm size.
Second, we specify the coefficients of the input-output table as a function of the
balance sheet conditions of suppliers and buyers. When a firm inputs certain goods
into the production process, it makes a decision about how much to purchase from
large suppliers and small suppliers. It is often argued that large firms with easy
access to bank credit can distribute their credit to their small customers by way of
trade credit. This is the so-called redistributional view of trade credit.5 Furthermore, the buyer may prefer a supplier with a healthy balance sheet, to ensure the
delivery of intermediate goods. We can test these conjectures in our framework.
To preview our findings, we find that the lending attitude of financial institutions toward suppliers, a proxy for the balance sheet conditions of the financial sector, affected buyers’ input decisions in the late 1990s and the early 2000s,
when Japanese financial institutions suffered from excessive non-performing loans.
Specifically, in the lost decade the customer, irrespective of its size, preferred to
purchase intermediate inputs from those suppliers that faced an easier lending attitude, rather than from those facing a more severe lending attitude. We also find
that customers, irrespective of their size, increased purchase of intermediate inputs
from large suppliers when liquidiy of small suppliers was reduced, small suppliers
became increasingly dependent on debt and/or sales growth of large suppliers in5
See Meltzer (1960), Jaffee (1971), Ramey (1992), Petersen and Rajan (1997), McMillan and
Woodruff (1999), Nilsen (2002), De Haan and Sterken (2006), and Love et al. (2007) for evidence
on the validity of the redistributional view of trade credit in the U.S. and other countries. For the
Japanese evidence, see Takehiro and Ohkusa (1995), Ono (2001), Ogawa (2003c), Uesugi and Yamashiro (2004), Uesugi (2005), Fukuda et al. (2006), Taketa and Udel (2006), Uchida et al. (2006),
Tsuruta (2008), and Ogawa et al. (2009).
3
creased in the lost decade. To gauge the quantitative importance of our findings,
we conduct simulation exercises to establish the extent to which change in lending
attitude affects the output of each industry, via change in inter-industry transactions. We find that an easier lending attitude toward small suppliers increased the
output in the small firm sector, and reduced the output in the large firm sector. This
suggests that differential changes in lending attitude toward the large firm sector
and the small firm sector bring about distributional changes in intermediate inputs across sectors with different firm size, which in turns leads to non-negligible
changes in the sectoral outputs.
The paper is organized as follows. In Section 2 we discuss the determinants
of input-output structure theoretically. Section 3 derives the basic equation to be
estimated, and describes the data set we use. Section 4 interprets the estimation
results we obtained, and Section 5 presents the results of the simulation exercises.
Section 6 concludes this study.
2 Determinants of the Input-Output Structure: Theoretical Discussions
In traditional input-output analysis, the input-output coefficient is technically determined. Suppose that a firm has the following constant-returns-to-scale CobbDouglas production technology.
α M1
Y = ALαL K αK M1
α MN
· · · MN
,
(1)
4
where
Y : gross output,
L : labor,
K : capital,
Mi : intermediary input from the i-th industry (i = 1, . . . , N), and
αL , αK , αm1 , . . . , αmN : technology parameters with
αL + αK +
N
∑
α Mi = 1.
i=1
The firm determines the optimal ratio of intermediary inputs to gross output that
maximizes its profit(π) , defined as follows:
π = pY − wL − rK −
N
∑
p Mi M i ,
(2)
i=1
where
p : output price,
w : wage rate,
r : rental price of capital, and
P Mi : price of the i-th intermediary input.
The first-order conditions yield the following input demand function for intermediary goods:
p Mi M i
= α Mi (i = 1, 2, . . . , N).
pY
(3)
This equation shows that the input-output coefficients on value terms are simply
the technology parameters of the production function.
When a firm has the option to purchase the i-th intermediary input from two
suppliers, a large firm and a small firm, then we have to specify how the customer
5
determines the proportion of intermediary goods purchased from each supplier.
Three determinants affect the customer’s decision to purchase from large or small
suppliers. First, the firm can reduce the risk that the order placed for the intermediary inputs is not delivered as scheduled, by diversifying the orders from large and
small suppliers. The total amount of the i-th intermediary input necessary to attain
optimal production is rewritten from eq.(3) as
Mi∗ =
α Mi pY
.
p Mi
(4)
Given the optimal amount of the i-th intermediary input given by eq.(4), the firm
determines the proportion of intermediary goods that it orders from large and small
suppliers in a way that minimizes the expected loss from failing to attain the profitmaximizing level of intermediary input. Formally, the objective function of the
customer is written as:
[
]2
E Mi∗ − ãiL MiL − ãiS MiS .
(5)
where
MiL : amount ordered from large suppliers,
MiS : amount ordered from small suppliers,
ãiL : stochastic factor that affects realization of the order from large firms, and
ãiS : stochastic factor that affects realization of the order from small firms.
The idea underlying our formulation is as follows. The firm knows the optimal total
amount of intermediary goods and places orders with large and small suppliers.
However, it takes some time for the ordered goods to be delivered to the customer,
and there is always some possibility that the goods delivered will fall short of those
ordered, due to stochastic shocks. Therefore the customer has an incentive to lessen
the risk by diversifying the orders between large and small suppliers. Formally the
6
firm minimizes eq.(5) subject to the following constraint:
MiL + MiS = Mi∗ .
(6)
The first condition yields the following demand function for the i-th intermediary
input of large suppliers.
[ ]
2
MiL E [ãiL ] − E [ãiS ] + E ãiS − E [ãiL ãiS ]
[
]
=
.
Mi∗
E (ãiL − ãiS )2
(7)
The term E [ãiL ] − E [ãiS ] measures the difference in the mean of the stochastic
factors. Here we assume that E [α̃iL ] = E [α̃iS ]. Then eq.(7) is written simply as
miL =
σ2iS − σiS L
MiL
=
,
Mi∗
σ2iS + σ2iL − 2σiS L
(8)
where
σ2iS : variance of ãiS
σ2iL : variance of ãiL , and
σiS L : covariance between ãiS and ãiL .
Similarly, the demand function for the i-th intermediary input of small suppliers is
given by6
miS =
σ2iL − σiS L
MiS
=
.
Mi∗
σ2iS + σ2iL − 2σiS L
(9)
We can show that when σ2iS > σ2iL then miL > miS . In other words, if the delivery
uncertainty of a small supplier is larger than that of a large supplier, the proportion
purchased from the large supplier is larger than that from the small supplier.7
Comparative statics also enable us to obtain the following results:
∂miL
∂miL
< 0,
> 0,
2
∂σiL
∂σ2iS
∂miS
∂miS
< 0,
> 0.
2
∂σiS
∂σ2iL
6
(10)
It can( be shown) that, if the correlation coefficient between ãiS and ãiL (ρi ) satisfies the condition
σiS σiL
ρi < min
,
, then 0 < miL , miS < 1 .
σiL σiS
7
This proposition and the subsequent comparative statics results remain essentially valid without
the constraint (6).
7
A rise in the delivery uncertainty of one supplier, measured by the variance of ãiS
or ãiL , will reduce the proportion of purchase from that supplier, and will instead
increase the proportion purchased from the other supplier. These results suggest
that the degree of uncertainty about delivery is very important in determining the
degree of diversification of intermediate inputs between large and small suppliers.
Note that the degree of uncertainty about delivery depends crucially on the balance
sheet conditions of the suppliers. When one supplier’s balance sheet deteriorates,
then it is quite likely that the supplier will be forced to reduce production, perhaps
due to the unavailability of working capital, and thus cannot deliver the contracted
amount to its customers. Therefore, the customer has an incentive to increase its
purchases from the supplier with a healthier balance sheet.
Second, the customer may prefer purchase from large firms, since large suppliers have better access to credit, and hence can redistribute the credit they receive to
their customers by way of trade credit. This is the redistributional aspect of trade
credit. Note that the redistributional role of trade credit depends on the balance
sheet conditions of financial institutions, since credit availability, for both large
and small firms, is very much affected by the health of financial institutions.
Finally, the market structure of intermediate goods is an important factor in
determining the purchase pattern of intermediate inputs from large and small suppliers. When a market for intermediate goods is oligopolistic, purchase will be
heavily dependent on large suppliers. On the other hand, dependence on large suppliers will be lower in a competitive intermediate goods market. It should be noted
that the input-output coefficient in our context is no longer the parameter determined purely by production technologies. The input-output coefficient, say from a
large supplier in the i-th industry, is defined as
p Mi MiL
p Mi Mi MiL
MiL
=
·
= α Mi ·
.
pY
pY
Mi
Mi
(11)
The first term of the right hand side of eq.(11) is the conventional input-output co8
efficient, which is technologically given, but the second term of the right hand side
of eq.(11) depends upon economic factors, such as the balance sheet conditions of
suppliers and financial institutions, and the market structure of intermediate goods.
3 Model Specification and the Data Set Description
3.1 Model Specification
In our model the input-output coefficient has four dimensions: buyer, supplier, firm
size of buyer and firm size of supplier. We assume that the economy consists of N
industries. Consider the production structure of the small firm in the j-th industry
( j = 1, 2, . . . , N) . Suppose that the small firm in the j-th industry buys MiL, jS units
of input from the large firm in the i-th industry when it produces Y jS units of output.
Then the input-output coefficient (aiL, jS ) in value terms is defined as
aiL, jS ≡
p Mi MiL, jS
.
p j Y jS
(12)
The coefficient aiL, jS is a product of the input-output coefficient p Mi Mi, jS /p j Y jS ,
where Mi, jS is the total input from the i-th industry to the small firm of the j-th
industry, and the proportion of inputs purchased from the large supplier of the i-th
industry miL, jS ≡ p Mi MiL, jS /p Mi Mi, jS . The former is an exogenous parameter of
the Cobb-Douglas production technology, while the latter depends on economic
factors, as described in the previous section.
Now we make an econometric specification of the determinants of miL, jS . First,
it will be affected by the balance sheet conditions of the suppliers. Deterioration in
the balance sheet of the supplier may prevent the order placed from being delivered
as scheduled. This effect can be captured by the debt outstanding relative to real
activities of the large supplier and the small supplier of the i-th industry, which we
denote by DEBT iL and DEBT iS , respectively. A fall (rise) in DEBT iL (DEBT iS )
will increase miL, jS .
9
Liquidity is another balance sheet variable of the supplier that we consider.
When the supplier has abundant liquidity, production activities will be executed
smoothly and thus the order placed will be delivered without delay. We denote the
liquidity of the large supplier and the small supplier of the i-th industry by LIQiL
and LIQiS , respectively. An increase (decrease) in LIQiL (LIQiS ) will increase
miL, jS .
The redistributional view of trade credit implies that the bank credit that suppliers receive may be redistributed to their customers via trade credit. Therefore
the necessary condition for the redistributional view to hold is that the supplier
receives sufficient credit from financial institutions. We use the lending attitude
of financial institutions toward the supplier as a proxy for the availability of bank
credit. The lending attitude of financial institutions to large suppliers and small
suppliers of the i-th industry is denoted by LENDiL and LENDiS , respectively. An
increase (decrease) in LENDiL (LENDiS ) will increase miL, jS .
Sales growth might also affect the purchase pattern of intermediate inputs between large and small suppliers. Higher sales growth will warrant stable supply
of intermediate goods to customers. We denote the growth rate of sales of the
large suppliers and the small suppliers of the i-th industry by S GROWT HiL and
S GROWT HiS , respectively. A rise (fall) in S GROWT HiL (S GROWT HiS ) will
increase miL, jS .
The market structure of the supplier is an important factor in determining the
pattern of purchases from large and small suppliers. Market structure is captured
in this study by the dummy variables, as follows. In specifying the miL, jS equation, we add the dummy variable DU MiL, jS to represent individual effects (i, j =
1, 2, . . . , N). The variable DU MiL, jS takes unity for the pair of large supplier in the
i-th industry and small customer in the j-th industry, and zero elsewhere. Then the
∑
average industry effect of supplier is calculated simply as (1/N) Nj=1 γiL, jS , where
10
γiL, jS is the coefficient estimate of DU MiL, jS .
Lastly, we take the balance sheet conditions of the buyer into consideration.
The balance sheet variables are debt outstanding relative to real activities (DEBT jS ),
liquidity (LIQ jS ) and sales growth rate (S GROWT H jS ). We also add the lending attitude of financial institutions toward the small customer of the j-th industry
(LEND jS ) to the list of explanatory variables.
To sum up, the equation to be estimated is written as8
miL, jS = γ0 + γ1 LIQiL + γ2 LIQiS + γ3 DEBT iL + γ4 DEBT iS
+γ5 LENDiL + γ6 LENDiS + γ7 S GROWT HiL + γ8 S GROWT HiS
+γ9 LIQ jS + γ10 DEBT jS + γ11 LEND jS + γ12 S GROWT H jS
+
N ∑
N
∑
γiL, jS DU MiL, jS +
i=1 j=1
T
∑
λt Dt + iL, jS (i, j = 1, 2, . . . , N), (13)
t=1
where
Dt : time dummy, and
iL, jS : disturbance term.
The proportion of inputs purchased from the small supplier of the i-th industry
(miS , jS ) does not give any additional information, since miS , jS is linearly related
to miL, jS as 1 − miL, jS . Therefore we use only the input information from large
suppliers. As for the input customer, small customers and large customers may
respond differently to changes in the balance sheet conditions, and in the lending
attitude of financial institutions. Therefore eq.(13) is estimated separately for large
8
Time dummies are also added to the equation to account for the effects of the macro shocks
common to each industry, since we pool different panels of input-output tables.
11
customers. The equation to be estimated for large customers is written as
miL, jL = η0 + η1 LIQiL + η2 LIQiS + η3 DEBT iL + η4 DEBT iS
+η5 LENDiL + η6 LENDiS + η7 S GROWT HiL + η8 S GROWT HiS
+η9 LIQ jL + η10 DEBT jL + η11 LEND jL + η12 S GROWT H jL
+
N ∑
N
∑
ηiL, jL DU MiL, jL +
i=1 j=1
T
∑
µt Dt + iL, jL (i, j = 1, 2, . . . , N), (14)
t=1
3.2 Data Set Description
The proportion of inputs purchased from either large suppliers or small suppliers,
(mik, jl ; k, l = S , L ), is directly estimated by the scale-wise input-output tables compiled by the Applied Research Institute Japan. We use the input-output tables of
1980, 1985, 1990, 1995 and 2000. In these tables, the sectors in the manufacturing
industry are further divided into two sectors by firm size. In the original tables, the
number of sectors in manufacturing industry is 23, which are aggregated into 14
sectors in accordance with the sector classification in Financial Statements Statistics of Corporations (abbreviated as FSSC), compiled by the Ministry of Finance.9
Since we restrict the analysis to manufacturing industry, the total number of input coefficients used in our analysis is 1,960 (=(14 suppliers)×(14 customers)×(5
years)×(2 firm sizes)).10 In the estimation, we discard observations that report no
input from a certain industry, or negative values in the input-output tables. Also,
some sectors have negative input coefficients, mainly due to the treatment of waste
or by-products. We also eliminated these observations from the sample.
The distribution of the input coefficients (miL, jl ; l = S , L) and the related de-
9
The sector concordance between the Input-output tables and the Financial Statement Statistics
of Corporations is presented in Table A1 of the Data Appendix.
10
The total number of input coefficients is 3,920 (=1,960×2) but, as discussed above, the proportion of input purchased from small suppliers is linearly related to that from large suppliers. Therefore
the number of input coefficients used in our analysis is 1960.
12
scriptive statistics are presented in Table 1.11 The mean of miL, jl has remained relatively stable since 1985, irrespective of firm size. It ranges from 0.401 to 0.436.
The mode of distribution also remains unaltered over time, and is in the interval of
0.1 to 0.2, irrespective of firm size. The shape of the distribution is bimodal.
All the balance sheet variables are taken from FSSC.12 The FSSC data are on
a fiscal year basis, and we have the values at the beginning of period and at the end
of the period available for stock variables. To maintain the consistency of the data
frequency with the input-output tables, we use the stock variable at the beginning
of the period. The debt outstanding relative to real activities is measured in two
ways. One is the ratio of debt to sales (DEBT 1), and the other is the ratio of total
borrowings to sales (DEBT 2). The liquidity variable (LIQ) is defined as the ratio
of cash, deposits and securities in current assets to sales. The sales variable is
deflated by the output deflator of each industry reported in the Annual Report of
National Account (Cabinet Office of the Government of Japan).
The lending attitude of financial institutions comes from The Short-term Economic Survey of Enterprises or Tankan Survey, released by the Bank of Japan. The
original series is available quarterly, so we use annual averages.
Table A3 summarizes the balance sheet variables and the lending attitude thus
constructed, by firm size and industry for each year. It should be noted that the
variation in these variables over the whole sample is small relative to those of the
input coefficients. That is to say, the balance sheet variables of the i-th supplier
take the same value irrespective of its customers, and those of the j-th customer
takes the same value irrespective of its suppliers.
11
The original distribution of the input coefficients is shown in Table A2 in the Data Appendix.
Comparison of Table 1 and Table A2 reveals how many observations have been eliminated from the
sample, due to zero or negative inputs.
12
In the input-output tables, small firms are defined as those whose number of employees is less
than 300, while in the FSSC we define small firms as those whose equity capital is less than 100
million yen.
13
4 Estimation Results and their Implications
Table 2 shows the estimation results when DEBT 1 is used as the DEBT variable.
The estimation is conducted for the whole period, the period from 1980 to 1990
and the period of the lost decade (1995 and 2000). First we examine the estimation results for small customers. When the estimation is conducted for the whole
period, the debt-sales ratio of large suppliers has a significantly negative effect on
the proportion of purchases from large suppliers. An increase in the debt burden
on large suppliers prompts small customers to rely more on small suppliers, due
to increasing uncertainty about the delivery of intermediate inputs from large suppliers with a shaky balance sheet. In the lost decade period the debt-sales ratio of
small suppliers exerts a significantly positive effect on the proportion of purchase
from large suppliers. In other words, a rise in the debt-sales ratio of small suppliers induces small customers to depend more on large suppliers. We also find that
liquidity of small suppliers has negative effects on the proportion of purchase from
large suppliers, implying that fall of liquidity of small suppliers prompts small customers to rely more on large suppliers more abundant in liquidity. Furthermore,
in the lost decade period, the lending attitude of financial institutions toward large
(small) suppliers has a significantly positive (negative) effect on the proportion of
purchases from large suppliers. This result indicates that easing the lending attitude
toward large suppliers and / or tightening the lending attitude toward small suppliers raises the proportion of purchases from large suppliers by small customers.
This result is consistent with the redistributional view of trade credit. Lastly, we
find that higher sales growth of large suppliers, which warrants stable supply of
intermediate goods to customers, makes small customers more dependent on large
suppliers.13
13
Significantly positive coefficient of sales growth of small suppliers is a bit puzzling result to
interpret. It might suggest that purchase from small and large suppliers are complements rather than
substitutes.
14
Now we turn to the estimation results for large customers. When the estimation is conducted for the whole period, the debt-sales ratio of large suppliers has
a significantly negative effect on the proportion of purchases from large suppliers.
In the latter period we find that higher sales growth of large suppliers, lower liquidity of small suppliers and higher debt-sales ratio of small suppliers significantly
increase dependence on large suppliers. We also find that easier lending attitude
toward large suppliers and/or tighter lending attitude toward small suppliers increase dependence on large suppliers significantly. This result indicates that the
redistributional view of trade credit is valid, even for large firms in the lost decade.
It should be noted that the market structure of suppliers, shown in the bottom
panel of the table, is important, irrespective of the sample period and the type of
customer. The figures in the table measure the magnitude of the industry effect,
relative to the food products and beverages industry. Almost all the parameter
estimates are significantly positive. We observe large values for the petroleum and
coal products, electrical machinery and transport equipment industries.
Table 3 shows the estimation results when DEBT 2 is used as the DEBT variable. The results remain essentially unaltered. In the lost decade, the coefficient
estimate of the debt-sales ratio of small suppliers is significantly positive for small
customers . We find significantly positive (negative) effects of lending attitude toward large (small) suppliers on the proportion of purchases from large suppliers,
for both large and small customers. We also find that higher sales growth of large
suppliers and lower liquidity of small suppliers significantly increase dependence
on large suppliers, for both large and small customers. The market structure of
suppliers is also important for customers’ purchase behavior, irrespective of the
sample period and the type of customer.
15
5 The Impact of Balance Sheet Contagion on Sectoral Output by Inter-Industry Linkage: Simulation Analysis
The virtue of input-output analysis is that it enables us to evaluate quantitatively
to what extent an initial increase in final demand in one sector is propagated into
output in other sectors, and eventually in aggregate output. This is well known as
the multiplier effect. The inverse matrix of identity matrix minus input-coefficient
matrix plays a vital role in determining the magnitude of multipliers. In the previous sections, we showed that when firm size is taken into consideration in the
inter-industry transactions, the input-output coefficients are not technically determined constant, but depend on the balance sheet conditions of firms and financial
institutions. The upshot is that the multiplier effects are also affected by the balance sheet conditions of firms and financial institutions. Furthermore, change in the
balance sheet conditions also brings about sectoral reallocation of outputs through
substitution of intermediate inputs between large and small firm sector.
In this section we quantitatively evaluate to what extent sectoral outputs are
affected by change in the balance sheet conditions. Specifically, we conduct the
following simulation exercise. It has been often argued that small firms suffered
most in the credit crunch in the late 1990s in Japan. Figure 1 shows the difference
of the lending attitude toward large firms and small firms in 1995 by industry. Note
that the lending attitude is much easier toward large firms except for petroleum
and coal products. In particular the lending attitude is easier toward large firms by
more than 20 percentage points for textiles, fabricated metal products and precision
instruments.
We quantitatively evaluate the situation where the lending attitude toward small
firms gets easier. Specifically, we assume that the lending attitude of financial
institutions toward small firms in 1995 gets as easy as toward large firms across all
16
manufacturing industries, keeping the lending attitude toward large firms intact.
14
In this simulation, we adopt the estimated equations for the period 1995-2000
in Table 2, where DEBT 1 (Debt / Sales ratio) is used as the DEBT variable. The
impact of this scenario on sectoral output in 1995 is calculated in the following
steps. First we compute the input-output coefficient matrix of the base case in
1995, using the predicted values of miL, jS and miL, jL , from eqs.(13) and (14), by
substituting the historical values in 1995 into each explanatory variable.15 In other
words,
âiL, jS = bi, jS m̂iL, jS ,
âiS , jS = bi, jS (1 − m̂iL, jS ),
(15)
âiL, jL = bi, jL m̂iL, jL , and
âiS , jL = bi, jL (1 − m̂iL, jL ),
where
m̂iL, jS : predicted values of miL, jS in 1995 computed from eq.(13),
m̂iL, jL : predicted values of miL, jL in 1995 computed from eq.(14),
bi, jS =
p Mi Mi, jS
: actual ratio of input from the i-th industry to output of
p j Y jS
small firms in the j-th industry in 1995, and
bi, jL =
p Mi Mi, jL
: actual ratio of input from the i-th industry to output of
p j Y jL
large firms in the j-th industry in 1995.
Then we calculate the inverse matrix of I−(I−V) , where the elements of  matrix
are given by eq.(15) and V is a diagonal matrix where the diagonal elements are
the ratios of import to the domestic demand for the corresponding industries.16
14
Note that in this scenario the lending attitude of financial institutions toward small firms tightens
in petroleum and coal products.
15
For the predicted year, 1995, the mean absolute error of m̂iL, jl is 0.0206 for small firms (l = S )
and 0.0171 for large firms (l = L). In terms of the original input coefficients, âiL, jl = bi, jl m̂iL, jl used for
the simulation, the mean absolute errors are negligibly small: 0.00064 for small firms and 0.00049
for large firms.
16
The predicted m̂iL, jl (l = S , L) can exceed unity or take a negative value. This case is quite likely
17
In the next step we compute the input-output coefficient matrix under this scenario, by substituting the newly assumed values of the lending attitude variable
into m̂iL, jS and m̂iL, jL equation, with the other variables taking the same values
as before. We denote the input-output coefficient matrix thus calculated by Ã.
Change in sectoral outputs induced by the domestic final demand are the elements
[
]−1
of I − (I − V)Ã (I − V).
Change in sectoral outputs is composed of two parts. One is the change in
sectoral outputs due to the change in the input-output coefficient matrix. This part
is calculated as
[[
]−1 [
]−1 ] [
]
I − (I − V)Ã − I − (I − V)Â
(I − V)f d + e ,
(16)
where f d is the domestic final demand vector, including private consumption, private investment, inventory change, and government expenditure, and e is the vector
of export in 1995. This term reflects substitution of intermediate inputs between
small firms and large firms.
The other part is the change in sectoral output induced by a change in final demand. Note that change in the balance sheet conditions of firms and financial institutions might affect investment, important component of domestic final demand.17
This part is written as
[
]−1
I − (I − V)Ã (I − V)∆f d ,
(17)
where ∆f d is the change in domestic final demand in 1995 arising from the change
in balance sheet conditions.
Now we turn to quantitative evaluation of the scenario. The first column of
Table 4 shows the sectoral output before the change in the lending attitude of fiwhen actual miL, jl is very close to unity or zero, since our prediction is based on OLS with a fixed
effect model. Actually, we have 10 (m̂iL, jS > 1) and 1 (m̂iL, jS < 0) cases out of 179 observations for
small firms, and 10 ( m̂iL, jL > 1) and 2 ( m̂iL, jL < 0) cases out of 182 observations for large firms. In
these cases we replace them with 1 or 0.
17
For example, see Ogawa(2003b) for the effects of balance sheet conditions of firms and financial
institutions on corporate investment.
18
[
]−1 [
]
nancial institutions, calculated as I − (I − V)Â
(I − V)f d + e . The second column shows the sectoral output after the change in the lending attitude, calculated
[
]−1 [
]
as I − (I − V)Ã
(I − V)f d + e . The third column is the difference between the
second column and the first one. The figures in the third column represent how
much the output of a certain industry changes when the lending attitude toward
small firms gets as easy as toward large firms, with the final demand being fixed.
As for the change in final demand, based on the investment function estimated
in Ogawa(2003b), easing lending attitude toward small firms in this scenario increases corporate investment of small firms by 682.6 billion yen. This increase
of investment is then allocated across industries, using the weights of the private
gross fixed capital formation by industry in 1995. The fourth column shows the
increase of sectoral outputs brought about by this increment of final demand. The
fifth column is the total change in sectoral output, sum of the third and the fourth
column. The sixth column shows the rate of change in sectoral output.
The table also shows the grand total of the figures over large firms in manufacturing industries, and that over small firms in manufacturing industries. The former
corresponds to the total increase in the output of large firms in all manufacturing
industries, while the latter corresponds to that of the output of small firms in all
manufacturing industries.
The third column of Table 4 shows that the output of small manufacturing firms
increases by 8,310.5 billion yen and that of large manufacturing firms decreases by
8,986.6 billion yen. The output of the manufacturing firms as a whole decreases
by 676.1 billion yen. This indicates that intermediate inputs purchased from large
manufacturing firms is substituted by those from small manufacturing firms that
now face lending attitude as favorable as large firms.
Induced by increase of final demand, the output of small and large manufacturing firms is raised by 281.6 billion yen and 280.4 billion yen, respectively. Com-
19
parison of the third column with the fourth column shows that substitution effects
dominate the multiplier effects. Consequently the output of small manufacturing
firms increases by 8,592.1 billion yen, while that of large manufacturing firms decreases by 8,706.2 billion yen. Change in the output varies across industries. In
large manufacturing firms the change is notably large for textile (-86.9%) and fabricated metal products (-20.6%). On the other hand, in small manufacturing firms the
change is large for iron and steel (20.8%), non-ferrous metals (17.9%), transport
equipment (14.8%) and textile (14.3%).
Figure 2 shows the scatter diagram of the rate of change in output of small
manufacturing firms and the change in lending attitude of financial institutions
toward small firms across industries. We observe positive correlation of the rate of
change in lending attitude with the rate of change in output. In fact the correlation
coefficient is 0.41.
Our approach is contrasted with the conventional one. In the conventional
approach favorable change in lending attitude toward small firms is analyzed as
follows. As is shown above, favorable change in lending attitude toward small
firms creates 682.6 billion yen increase of corporate investment, which is allocated across industries as additional final demand, using the weights of the private
gross fixed capital formation by industry in 1995. Then the multiplier is calculated
based on the input-output coefficient matrix without taking account of the effects
of change in lending attitude. Change in output thus calculated is shown in the seventh column of Table 4. Comparison of the fifth column and the seventh column
shows that the change in output is overestimated for large manufacturing firms and
underestimated for small manufacturing firms in the conventional approach. This
is due to omission of substitution effects of intermediate inputs from large manufacturing firms to small manufacturing firms in the conventional approach. The
total multiplier is also quite different between the two approaches. The multiplier
20
in our approach is 0.926 (= 632.1 / 682.6), while it is 1.894 (= 1,292.9 / 682.6) in
the conventional case.
The simulation results above indicate quantitative importance of substitution
effects of intermediate inputs between large and small manufacturing firms. It also
hints that output of small firm sector increases to a large extent simply by easing
lending attitude toward them without any increase in final demand.
6 Concluding Remarks
This paper proposed a novel approach to investigating the propagation mechanism
of balance sheet deterioration in financial institutions and firms, by extending the
conventional input-output analysis. The direction of extension is twofold. One is
the use of input-output tables that are classified by firm size for the manufacturing
sector. This adds another dimension to the inter-industry structure: the transactional relationship between firms of different sizes. The other links the input-output
tables with the balance sheet conditions of financial institutions and firms, and this
enables us to analyze customers’ decision making in allocating input purchases
between large and small suppliers.
By pooling the Japanese input-output tables, classified by firms, for 1980,
1885, 1990, 1995 and 2000, we explored the determinants of the purchase of intermediate goods from large and small suppliers. We found that the lending attitude
of financial institutions affected customers’ input decisions from the late 1990s to
the early 2000s.
Based on the estimation results, we conducted simulation exercises to evaluate
quantitatively the extent to which the change in the balance sheet conditions of
financial institutions that was favorable to small firms affected the sectoral outputs.
We found that the output increased for small firms and declined for large firms.
The change in output is non-negligible, about 5.5% of the initial output of each
21
sector. In particular it exceeded 20% in textile, iron and steel and fabricated metal
products. This suggests that a change in the balance sheet conditions of financial
institutions generates a non-negligible distributional change in output among firms
of different sizes.
References
Basu, S.(1995). ”Intermediate Goods and Business Cycles: Implications for Productivity and Welfare,”American Economic Review 85, pp.512-531.
Bergin, P.R. and R.C. Feenstra (2000). ”Staggered Price Setting, Translog Preference, and Endogenous Persistence,” Journal of Monetary Economics 45,
pp.657-680.
Boissay, F. (2006). ”Credit Chains and the Propagation of Financial Distress,”
European Central Bank Working Paper Series No.573.
Caballero,R., Kashyap, N., and T. Hoshi (2008). ”Zombie Lending and Depressed
Restructuring in Japan,” American Economic Review 98, pp. 1943 - 77
De Haan, L. and E. Sterken (2006), ”The impact of monetary policy on the financing behaviour of firms in the Euro area and the UK,” European Journal
of Finance 12, No.5, pp.401-420.
Dupor, B. (1999). ”Aggregation and Irrelevance in Multi-Sector Models,” Journal
of Monetary Economics 43, pp.391-409.
Fukuda, S., Kasuya, M. and K. Akashi (2006). ”The Role of Trade Credit for
Small Firms: An Implication from Japan’s Banking Crisis,” Bank of Japan
Working Paper Series, No. 06-E-18.
Hertzel, M. G., Z. Li, M.S. Officer and K. J. Rodgers (2008). ”Inter-Firm Link22
ages and the Wealth Effects of Financial Distress along the Supply Chain,”
Journal of Financial Economics 87, pp.374-387.
Hornstein, A. and J. Praschnik (1997). ”Intermediate Inputs and Sectoral Comovement in the Business Cycle,” Journal of Monetary Economics 40,
pp.573-595.
Horvath, M. (1998). ”Cyclicality and Sectoral Linkages: Aggregate Fluctuations
from Independent Sector-Specific Shocks,” Review of Economic Dynamics
1, pp.781-808.
Horvath, M. (2000). ”Sectoral Shocks and Aggregate Fluctuations,” Journal of
Monetary Economics 45, pp.69-106.
Huang, K.X.D. and Z. Liu (2001). ”Production Chains and General Equilibrium
Aggregate Dynamics,” Journal of Monetary Economics 48, pp.437-462.
Jaffee, D. (1971). Credit Rationing and the Commercial Loan Market, John Wiley
and Sons.
Kiyotaki, N. and J. Moore (1997). ”Credit Chains,”mimeographed.
Kiyotaki, N. and J. Moore (2002). ”Balance-Sheet Contagion,” American Economic Review 92, pp.46-50.
Kobayashi, K. and M. Inaba (2002).”Japan’s Lost Decade and the Complexity
Externality,” RIETI Discussion Paper Series 02-E-004.
Lang, L. and R. Stulz (1992). ”Contagion and Competitive Intra-Industry Effects
of Bankruptcy Announcements,” Journal of Financial Economics 32, pp.4560.
23
Long, J. and C. Plosser (1983). ”Real Business Cycles,” Journal of Political Economy 91, pp.39-69.
Love. I., Preve, L.A., V. Sarria-Allende (2007). ”Trade Credit and Bank Credit:
Evidence from Recent Financial Crises,” Journal of Financial Economics
83, pp.453-469.
McMillan, J. and C. Woodruff (1999). ”Interfirm Relationships and Informal
Credit in Vietnam,” The Quarterly Journal of Economics 114, pp.1285-1320.
Meltzer, A. (1960). ”Mercantile Credit, Monetary Policy, and the Size of Firms,”
Review of Economics and Statistics 42, pp.419-436.
Nilsen, J. H. (2002). ”Trade Credit and the Bank Lending Channel,” Journal of
Money, Credit and Banking 34, pp.226-253.
Nishimura, K. G., Nakajima, T. and K. Kiyota (2005). ”Does the Natural Selection Mechanism Still Work in Severe Recessions? Examination of the
Japanese Economy in the 1990s,” Journal of Economic Behavior and Organization 58, pp.53-78.
Ogawa, K. (2003a). ”Financial Distress and Employment: The Japanese Case in
the 90s,” NBER Working Paper 9646.
Ogawa, K. (2003b). ”Financial Distress and Corporate Investment: The Japanese
Case in the 1990s,” Discussion Paper No.584, Institute of Social and Economic Research, Osaka University.
Ogawa,K. (2003c). Daifukyo no Keizaibunseki (Economic Analysis of the Great
Recessions, Nippon Keizai Shinbunsha, Tokyo (in Japanese).
Ogawa, K., Sterken, E. and I. Tokutsu (2009). ”Redistributional View of Trade
24
Credit Revisited: Evidence from Micro Data of Japanese Small Firms,” RIETI Discussion Paper Series 09-E-029.
Ono, M. (2001). ”Determinants of Trade Credit in the Japanese Manufacturing
Sector,” Journal of the Japanese and International Economies 15, No.2 pp.160-177.
Petersen, M.A. and R.G. Rajan (1997). ”Trade Credit: Theories and Evidence,”
The Review of Financial Studies 10, pp.661-691.
Raddatz, C. (2008). ”Credit Chains and Sectoral Comovement: Does the Use of
Trade Credit Amplify Sectoral Shocks?” The World Bank Policy Research
Working Paper 4525.
Ramey, V. (1992). ”Source of Fluctuations in Money: Evidence from Trade
Credit,” Journal of Monetary Economics 30, pp.171-193.
Shea, J. (2002). ”Complementarities and Comovement,” Journal of Money, Credit
and Banking 34, pp.412-433.
Shimoda, M., Fujikawa, K., and T. Watanabe (2005). ”Industrial Structure and
Size of Enterprises in Japan,” Sangyo Renkan (Input-Output Analysis) 13,
No.3, pp.52-65 (in Japanese).
Takehiro, R. and Y. Ohkusa (1995). ”Kigyokan Sinyo no Paneru Suitei (Panel
Estimation of Trade Credit),” JCER Economic Journal 28, pp.53-75 (in
Japanese).
Taketa,K. and G.F. Udell (2006). ”Lending Channels and Financial Shocks: The
Case of SME Trade Credit and the Japanese Banking Crisis,” IMES Discussion Paper, Bank of Japan, No. 2006-E-29.
25
Tsuruta, D. (2007), ”Credit contagion and trade credit supply: evidence from
small business data in Japan,” RIETI discussion paper series, 07-E-043.
Tsuruta, D. (2008). ”Bank Information Monopoly and Trade Credit: Do Only
Banks Have Information of Small Businesses?” Applied Economics 40, pp.
981- 996.
Uchida, H., Udell, G. F. and W. Watanabe (2006). ”Are Trade Creditors Relationship Lenders?” RIETI Discussion Paper Series 06-E-026.
Uesugi, I. and G. M. Yamashiro (2004). ”How Trade Credit Differs from Loans:
Evidence from Japanese Trading Companies,” RIETI Discussion Paper Series 04-E-028.
Uesugi, I. (2005). ”Kigyo-kan-Shinyo to Kinyu Kikan Kariire ha Daitai-teki ka
(Are Trade Credit and Financial Institution Borrowing Substitutes?),” JCER
Economic Journal 52, pp.19-43 (in Japanese).
26
(14) Miscellaneous manufacturing
18
(13) Precision instruments
21.5
(12) Transport equipment
14.2
(11) Electrical machinery
10.1
(10) Machinery
16.7
(9) Fabricated metal products
22.8
(8) Non−ferrous metals
14.5
(7) Iron and steel
8.5
(6) Non−metallic mineral products
(5) Petroleum and coal products
16.3
−8.3
(4) Chemicals
3.6
(3) Pulp, paper & paper products
3.2
(2) Textiles
21.5
(1) Food products and beverages
−30
−20
16.3
−10
0
10
20
Figure 1: Difference in Lending Attitude of Financial Institutions between Large Firms
and Small Firms: 1995
27
30
25
20
15
10
5
−5
0
Rate of change in output (%)
−10
−5
0
5
10
15
20
25
Change in the lending attitude toward small firms
Figure 2: Relationship between the Rate of Change in Output and the Change in Lending
Attitude across Industries: Small Manufacturing Firms
28
Table 1: Distribution of Normalized Input Coefficients by Year
(1)
1980
(2)
1985
(3)
1990
5
12
26
16
25
16
12
7
17
3
8
18
5
15
31
22
17
14
18
14
7
2
1
16
2
20
34
16
18
16
21
11
10
4
13
1
165
0.139
0.463
0.315
162
0.130
0.406
0.294
166
0.018
0.401
0.284
(4)
1995
(5)
2000
(6)
Total
2
16
35
20
18
18
27
17
8
1
15
2
2
27
30
19
14
21
17
15
10
6
15
4
16
90
156
93
92
85
95
64
52
16
52
41
179
0.022
0.416
0.278
180
0.033
0.414
0.302
852
0.067
0.420
0.295
Small firms
miL, jS = 0.0
0.0 < miL, jS ≤ 0.1
0.1 < miL, jS ≤ 0.2
0.2 < miL, jS ≤ 0.3
0.3 < miL, jS ≤ 0.4
0.4 < miL, jS ≤ 0.5
0.5 < miL, jS ≤ 0.6
0.6 < miL, jS ≤ 0.7
0.7 < miL, jS ≤ 0.8
0.8 < miL, jS ≤ 0.9
0.9 < miL, jS < 1.0
miL, jS = 1.0
Total
Fraction of 0 or 1 coefficients
Mean
Standard deviation
Large firms
miL, jL = 0.0
0.0 < miL, jL ≤ 0.1
0.1 < miL, jL ≤ 0.2
0.2 < miL, jL ≤ 0.3
0.3 < miL, jL ≤ 0.4
0.4 < miL, jL ≤ 0.5
0.5 < miL, jL ≤ 0.6
0.6 < miL, jL ≤ 0.7
0.7 < miL, jL ≤ 0.8
0.8 < miL, jL ≤ 0.9
0.9 < miL, jL < 1.0
miL, jL = 1.0
Total
Fraction of 0 or 1 coefficients
Mean
Standard deviation
5
12
25
15
14
19
16
8
23
3
6
20
5
14
27
23
14
17
14
22
11
3
1
17
4
18
29
17
18
18
21
15
10
4
13
2
4
17
29
19
18
20
24
18
13
3
13
4
4
26
24
13
20
22
20
14
12
7
15
4
22
87
134
87
84
96
95
77
69
20
48
47
166
0.151
0.488
0.317
168
0.131
0.435
0.299
169
0.036
0.416
0.283
182
0.044
0.436
0.283
181
0.044
0.428
0.303
866
0.080
0.440
0.297
29
Table 2: Estimated Results for Small Customers: DEBT 1
(1)
1980 - 2000
(2)
1980 - 1990
(3)
1995 - 2000
LIQiL
LIQiS
0.2534
-0.0832
(1.34)
(0.77)
0.6207
-0.1496
(1.81)*
(0.42)
-0.6270
-0.3456
(1.21)
(3.36)***
DEBT 1iL
DEBT 1iS
-0.1476
0.0224
(2.63)***
(0.36)
0.0184
-0.4412
(0.19)
(2.88)***
-0.0203
0.3112
(0.11)
(3.41)***
LENDiL
LENDiS
0.0004
-0.0008
(0.50)
(1.07)
0.0002
0.0006
(0.14)
(0.44)
0.0047
-0.0084
(4.00)***
(4.95)***
(2.41)**
(1.37)*
0.0283
0.0401
(0.38)
(1.11)
0.3626
0.1349
(2.52)**
(5.20)***
S GROWT HiL
S GROWT HiS
0.1482
0.0266
LIQ jS
DEBT 1 jS
LEND jS
S GROWT H jS
0.0393
-0.0612
-0.0007
0.0012
(0.38)
(0.96)
(1.48)
(0.06)
-0.0683
0.0563
-0.0009
-0.0352
(0.22)
(0.39)
(0.98)
(0.99)
-0.0259
0.0353
0.0002
0.0007
(0.33)
(0.65)
(0.15)
(0.03)
D1985
D1990
D1995
D2000
Constant term
-0.0175
-0.0311
-0.0232
-0.0402
0.1566
(0.51)
(1.33)
(0.64)
(1.48)
(3.06)***
-0.0200
-0.0202
0.2151
(0.36)
(0.49)
(2.50)**
-0.0358
-0.0296
(1.58)
(0.28)
Textiles
Pulp and paper products
Chemicals
Petroleum and coal
Non-metallic mineral
Iron and steel
Non-ferrous metals
Fabricated metal products
Machinery
Electrical machinery
Transport equipment
Precision instruments
Miscellaneous
0.1178
0.1761
0.4336
0.9799
0.2963
0.5384
0.5016
0.1074
0.2292
0.5986
0.6044
0.4721
0.1247
(3.62)***
(4.51)***
(14.7)***
(26.1)***
(9.92)***
(15.3)***
(16.5)***
(4.41)***
(10.8)***
(24.6)***
(27.4)***
(19.5)***
(5.60)***
0.1051
0.0821
0.3770
0.9318
0.2584
0.4612
0.4404
0.0662
0.1767
0.3727
0.4809
0.4338
0.0875
(2.46)**
(1.34)
(8.35)***
(17.6)***
(5.77)***
(8.66)***
(9.63)***
(1.94)*
(5.53)***
(11.0)***
(19.0)***
(11.3)***
(2.92)***
-0.0062
0.1116
0.6229
1.1744
0.2453
0.4396
0.4093
0.0533
0.2548
0.6116
0.5711
0.3942
0.1289
(0.06)
(1.01)
(11.4)***
(15.6)***
(3.48)***
(4.69)***
(4.46)***
(1.37)
(8.33)***
(19.8)***
(24.6)***
(12.7)***
(4.75)***
R̄2 / Se
N
0.9232
852
0.0817
0.9263
493
0.0810
0.9636
359
0.0472
The figures in parentheses are the t-values in absolute value. Asterisks *, **, and *** indicate that
the corresponding coefficients are significant at the 10%, 5% and 1% level, respectively. R̄2 , S e,
and N are the coefficients of determination adjusted for the degree of freedom, standard error of the
regression, and the number of observations, respectively.
30
Table 2: (continued) Estimated Results for Large Customers: DEBT 1
(1)
1980 - 2000
(2)
1980 - 1990
(3)
1995 - 2000
LIQiL
LIQiS
0.0358
-0.0874
(0.18)
(0.79)
0.0846
-0.0629
(0.24)
(0.17)
-0.6151
-0.1808
(1.57)
(2.35)**
DEBT 1iL
DEBT 1iS
-0.1271
0.0164
(2.20)**
(0.25)
-0.1285
-0.2111
(1.30)
(1.31)
0.1334
0.1940
(0.95)
(2.80)***
LENDiL
LENDiS
-0.0003
-0.0001
(0.43)
(0.10)
-0.0005
0.0005
(0.45)
(0.32)
0.0038
-0.0047
(4.31)***
(3.61)***
(0.94)
(0.67)
0.2870
0.0513
(2.62)***
(2.61)***
S GROWT HiL
S GROWT HiS
0.1735
0.0072
(2.73)***
(0.36)
0.0731
0.0255
LIQ jL
DEBT 1 jL
LEND jL
S GROWT H jL
0.1067
-0.0530
-0.0004
-0.1539
(0.57)
(0.96)
(0.92)
(2.39)**
0.3514
0.0589
-0.0001
-0.1792
(1.11)
(0.63)
(0.09)
(2.35)**
D1985
D1990
D1995
D2000
Constant term
0.0105
-0.0438
-0.0112
-0.0232
0.2241
(0.25)
(1.85)*
(0.28)
(0.79)
(4.08)***
0.0000
-0.0347
(0.00)
(0.88)
Food and beverages
Textiles
Pulp and paper products
Chemicals
Petroleum and coal
Non-metallic mineral
Iron and steel
Non-ferrous metals
Fabricated metal products
Machinery
Electrical machinery
Transport equipment
Precision instruments
Miscellaneous
0.0830
0.1920
0.4225
0.9220
0.3582
0.4147
0.5053
0.0912
0.2574
0.5994
0.6231
0.4837
0.1299
(2.48)**
(5.16)***
(13.8)***
(23.5)***
(11.6)***
(12.4)***
(14.6)***
(3.62)***
(10.8)***
(24.2)***
(27.6)***
(19.4)***
(5.63)***
R̄2 / Se
N
0.9173
866
0.0855
0.1400
-0.0486
0.0001
0.0156
(0.64)
(0.63)
(0.20)
(0.17)
0.2732
(3.23)***
-0.0109
-0.0434
(0.67)
(0.49)
0.0782
0.1604
0.3985
0.8763
0.3508
0.4026
0.4748
0.0692
0.2405
0.4582
0.5767
0.4666
0.1014
(1.79)*
(2.72)***
(8.50)***
(15.7)***
(7.57)***
(7.80)***
(9.15)***
(1.95)*
(6.67)***
(11.9)***
(20.7)***
(11.7)***
(3.27)***
-0.0627
0.0759
0.4950
1.0213
0.2690
0.2923
0.3664
0.0377
0.2479
0.6328
0.5841
0.4341
0.1260
(0.81)
(0.97)
(12.0)***
(17.9)***
(5.06)***
(4.54)***
(4.91)***
(1.29)
(9.88)***
(27.4)***
(35.1)***
(18.6)***
(6.27)***
0.9191
503
0.0856
0.9849
363
0.0360
The figures in parentheses are the t-values in absolute value. Asterisks *, **, and *** indicate that
the corresponding coefficients are significant at the 10%, 5% and 1% level, respectively. R̄2 , S e,
and N are the coefficients of determination adjusted for the degree of freedom, standard error of the
regression, and the number of observations, respectively.
31
Table 3: Estimated Results for Small Customers: DEBT 2
(1)
1980 - 2000
(2)
1980 - 1990
(3)
1995 - 2000
LIQiL
LIQiS
0.1985
-0.0606
(1.05)
(0.57)
0.6293
-0.3024
(1.80)*
(0.89)
-0.3026
-0.3149
(0.75)
(3.23)***
DEBT 2iL
DEBT 2iS
-0.1377
-0.0307
(1.82)*
(0.44)
0.0298
-0.5082
(0.27)
(2.51)**
-0.2586
0.1793
(1.65)
(1.93)**
LENDiL
LENDiS
0.0003
-0.0009
(0.46)
(1.13)
0.0000
0.0003
(0.03)
(0.24)
0.0041
-0.0076
(3.90)***
(4.52)***
(2.15)**
(1.73)*
0.0567
0.0361
(0.78)
(1.01)
0.3623
0.1123
(2.81)***
(4.61)***
S GROWT HiL
S GROWT HiS
0.1311
0.0335
LIQ jS
DEBT 2 jS
0.0428
-0.0794
(0.42)
(1.14)
0.0033
-0.0354
(0.01)
(0.19)
-0.0231
0.0439
(0.29)
(0.80)
LEND jS
S GROWT H jS
-0.0007
0.0023
(1.49)
(0.12)
-0.0009
-0.0314
(1.00)
(0.88)
0.0002
-0.0020
(0.24)
(0.09)
D1985
D1990
D1995
D2000
Constant term
-0.0112
-0.0282
-0.0149
-0.0364
0.1315
(0.32)
(1.17)
(0.40)
(1.26)
(2.94)***
0.0016
0.0075
0.1946
(0.03)
(0.16)
(2.58)***
-0.0313
0.0602
(1.37)*
(0.86)
Food and beverages
Textiles
Pulp and paper products
Chemicals
Petroleum and coal
Non-metallic mineral
Iron and steel
Non-ferrous metals
Fabricated metal products
Machinery
Electrical machinery
Transport equipment
Precision instruments
Miscellaneous
0.1016
0.1525
0.4127
0.9741
0.2786
0.4976
0.4835
0.0899
0.2120
0.5795
0.5895
0.4604
0.1118
(3.27)**
(4.01)***
(14.7)***
(23.3)***
(10.4)***
(17.8)***
(15.7)***
(3.92)***
(10.8)***
(22.7)***
(27.2)***
(19.2)***
(5.07)***
0.0573
0.0312
0.3328
0.8894
0.2050
0.4222
0.4062
0.0294
0.1370
0.3504
0.4553
0.4059
0.0498
(1.41)
(0.51)
(7.19)***
(14.1)***
(4.89)***
(10.1)***
(8.46)***
(0.87)
(4.44)***
(9.44)***
(18.4)***
(10.2)***
(1.57)
0.1166
0.2253
0.6138
1.2139
0.3256
0.5103
0.5038
0.0982
0.2779
0.6073
0.5800
0.4201
0.1543
(2.11)**
(3.26)***
(11.1)***
(18.0)***
(9.41)***
(13.6)***
(9.04)***
(3.87)***
(9.58)***
(15.6)***
(25.6)***
(12.6)***
(5.90)***
R̄2 /S e
0.9228
0.0819
0.9258
0.0814
0.9728
0.0478
N
852
493
359
The figures in parentheses are the t-values in absolute value. Asterisks *, **, and *** indicate that
the corresponding coefficients are significant at the 10%, 5% and 1% level, respectively. R̄2 , S e,
and N are the coefficients of determination adjusted for the degree of freedom, standard error of the
regression, and the number of observations, respectively.
32
Table 3: (continued) Estimated Results for Large Customers: DEBT 2
(1)
1980 - 2000
(2)
1980 - 1990
(3)
1995 - 2000
LIQiL
LIQiS
-0.0245
-0.0687
(0.13)
(0.63)
0.0885
-0.2802
(0.24)
(0.79)
-0.1108
-0.1875
(0.36)
(2.58)**
DEBT 2iL
DEBT 2iS
-0.1064
-0.0244
(1.37)
(0.34)
-0.0824
-0.1678
(0.73)
(0.79)
-0.1111
0.0711
(0.94)
(1.00)
LENDiL
LENDiS
-0.0004
-0.0001
(0.54)
(0.08)
-0.0006
0.0004
(0.46)
(0.26)
0.0031
-0.0043
(3.79)***
(3.35)***
(2.52)**
(0.63)
0.0717
0.0315
(0.95)
(0.84)
0.3434
0.0361
(3.48)***
(1.95)*
0.3420
0.0894
(1.09)
(0.77)
0.0570
-0.0010
(0.30)
(0.01)
0.0003
0.0009
(0.65)
(0.01)
(0.54)
(0.71)
S GROWT HiL
S GROWT HiS
0.1583
0.0127
LIQ jL
DEBT 2 jL
0.0776
-0.0284
(0.42)
(0.39)
LEND jL
S GROWT H jL
-0.0004
-0.1611
(0.91)
(2.51)**
-0.0001
-0.1738
(0.10)
(2.31)**
D1985
D1990
D1995
D2000
Constant term
0.0165
-0.0433
-0.0103
-0.0273
0.1936
(0.39)
(1.76)*
(0.25)
(0.90)
(3.94)***
0.0025
-0.0308
(0.04)
(0.68)
Food and beverages
Textiles
Pulp and paper products
Chemicals
Petroleum and coal
Non-metallic mineral
Iron and steel
Non-ferrous metals
Fabricated metal products
Machinery
Electrical machinery
Transport equipment
Precision instruments
Miscellaneous
0.0644
0.1679
0.4005
0.9114
0.3396
0.3798
0.4835
0.0745
0.2416
0.5831
0.6101
0.4733
0.1180
(2.03)**
(4.64)***
(13.9)***
(21.1)***
(12.3)***
(14.3)***
(13.9)***
(3.16)***
(10.9)***
(22.5)***
(27.5)***
(19.1)***
(5.19)***
R̄2 /S e
N
0.9169
866
0.0857
0.2203
(2.90)***
-0.0088
0.0436
0.0385
0.1194
0.3668
0.8646
0.3078
0.3535
0.4477
0.0424
0.2069
0.4408
0.5559
0.4498
0.0793
(0.93)
(2.03)**
(7.62)***
(13.2)***
(7.14)***
(8.58)***
(8.22)***
(1.21)
(5.88)***
(10.4)***
(20.2)***
(10.9)***
(2.43)**
0.0683
0.1973
0.4959
1.0730
0.3525
0.3852
0.4822
0.0773
0.2710
0.6186
0.5945
0.4441
0.1418
0.9182
503
0.0860
0.9844
363
(1.53)
(4.11)***
(12.0)***
(21.3)***
(13.9)***
(15.1)***
(10.7)***
(4.22)***
(11.4)***
(21.2)***
(36.7)***
(18.0)***
(7.45)***
0.0365
The figures in parentheses are the t-values in absolute value. Asterisks *, **, and *** indicate that
the corresponding coefficients are significant at the 10%, 5% and 1% level, respectively. R̄2 , S e,
and N are the coefficients of determination adjusted for the degree of freedom, standard error of the
regression, and the number of observations, respectively.
33
Table 4: Effect on Sectoral Output
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
(25)
(26)
(27)
(28)
(29)
(30)
(31)
(32)
(33)
(34)
(35)
(36)
(37)
(38)
(39)
(40)
(41)
(42)
(43)
(44)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
†1
†2
(2)−(1)
†3
(3)+(4)
(5) / (1)
†4
Public administration
Unclassified
15,808.3
1,660.7
8,460.0
30,395.6
423.7
3,626.0
2,920.5
6,498.4
13,385.1
12,405.1
9,921.9
566.8
1,787.5
7,908.2
15,350.1
4,769.3
3,705.8
2,643.2
3,669.2
12,042.9
13,820.6
14,657.0
36,428.4
13,949.2
33,616.1
8,171.9
1,786.8
2,023.8
11,390.8
36,211.3
88,149.9
26,462.5
52,112.4
50,212.3
100,521.3
55,666.3
37,647.7
79,222.4
100,351.7
7,555.7
15,913.1
1,653.3
8,068.0
30,773.7
55.5
4,138.0
2,735.6
6,606.6
13,036.2
12,688.0
9,937.6
525.8
1,480.2
8,229.6
14,390.2
5,746.1
3,227.7
3,109.9
2,902.5
12,807.2
13,134.7
15,329.2
35,421.7
14,839.4
32,096.8
9,363.4
1,713.2
2,096.8
9,480.0
37,925.4
88,150.9
26,439.5
52,126.0
50,250.1
100,526.4
55,669.6
37,613.5
79,192.6
100,293.3
7,559.8
104.9
-7.4
-392.1
378.2
-368.2
511.9
-184.9
108.2
-348.9
282.9
15.7
-40.9
-307.3
321.4
-959.8
976.9
-478.1
466.7
-766.7
764.2
-685.8
672.2
-1,006.6
890.2
-1,519.4
1,191.5
-73.6
73.0
-1,910.7
1,714.1
1.0
-23.0
13.6
37.8
5.1
3.2
-34.2
-29.9
-58.3
4.1
6.2
2.6
1.0
5.1
0.0
5.7
3.1
7.4
8.2
9.4
9.5
1.7
3.0
24.0
35.9
15.3
7.6
6.1
9.4
35.6
51.6
62.4
78.4
35.2
59.1
16.7
3.1
4.9
10.5
52.1
323.9
20.4
64.6
62.6
44.1
49.4
56.0
68.9
21.6
9.3
111.0
-4.8
-391.1
383.3
-368.2
517.6
-181.8
115.6
-340.7
292.3
25.1
-39.2
-304.3
345.4
-923.9
992.1
-470.5
472.8
-757.3
799.8
-634.3
734.6
-928.2
925.4
-1,460.2
1,208.2
-70.5
77.9
-1,900.3
1,766.2
324.9
-2.6
78.1
100.4
49.2
52.6
21.8
39.0
-36.8
13.3
0.7
-0.3
-4.6
1.3
-86.9
14.3
-6.2
1.8
-2.5
2.4
0.3
-6.9
-17.0
4.4
-6.0
20.8
-12.7
17.9
-20.6
6.6
-4.6
5.0
-2.5
6.6
-4.3
14.8
-3.9
3.8
-16.7
4.9
0.4
0.0
0.1
0.2
0.0
0.1
0.1
0.0
0.0
0.2
6.1
2.6
1.0
5.1
0.5
4.9
3.4
7.3
8.6
9.1
9.5
1.8
3.7
23.3
38.5
12.6
8.7
5.0
11.1
33.9
54.1
60.0
80.8
33.1
61.9
14.5
3.3
4.7
13.5
49.4
323.9
20.4
64.6
62.5
44.1
49.4
56.1
68.9
21.7
9.2
Large manufacturing
Small manufacturing
Manufacturing total
Industry total
156,666.5
155,868.6
312,535.1
927,906.4
147,679.9
164,179.1
311,859.1
927,247.1
-8,986.6
8,310.5
-676.1
-659.3
280.4
281.6
562.0
1,291.3
-8,706.2
8,592.1
-114.1
632.1
-5.6
5.5
0.0
0.1
298.5
264.8
563.3
1,292.9
Agriculture
Mining
Food and beverages
Textile
Pulp and paper
Chemicals
Petroleum and coal
Non-metallic mineral
Iron and steel
Non-ferrous metals
Fabricated metal
Machinery
Electrical machinery
Transport equipment
Precision instruments
Miscellaneous
Construction
Electricity
Wholesales and retails
Finance
Transportation
Service
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
(unit: billions of yen for columns (1) to (5), and (7); % for column (6))
[
]−1 [
]
†1 I − (I − V)Â
(I − V)f d + e
[
]−1 [
]
†2 I − (I − V)Ã
(I − V)f d + e
[
]−1
†3 I − (I − V)Ã
(I − V)∆f d
[
]−1
†4 I − (I − V)Â
(I − V)∆f d
34
Table A1. Sector Classification
Aggregated sectors in this study
Original sectors in input-output table
1
Food products and beverages
Food products
Beverages, tobacco and feeds
2
Textiles
Textiles
3
Pulp, paper and paper products
Pulp, paper and paper products
4
Chemicals
Chemicals
5
Petroleum and coal products
Petoleum products
Coal products
6
Non-metallic mineral products
Non-metalic mineral products
7
Iron and steel
Iron and steel
8
Non-ferrous metals
Non-ferrous metal
9
Fabricated metal products
Fabricated metal products
10
Machinery
Machinery
11
Electrical machinery, equipment and supplies
Electrical machinery, equipment and supplies
12
Transport equipment
Transport equipment
13
Precision instruments
Precision instruments
14
Miscellaneous manufacturing
Wearing apparel and clothing accessories
Wood and of wooden products
Furniture
Publishing and printing
Plastics products
Rubber products
Leather, fur products and miscellaneous leather products
Others
35
Table A2. Distribution of Input Coefficients by Year: Small Firms
(1)
1980
(2)
1985
(3)
1990
(4)
1995
(5)
2000
(6)
Total
aiL, jS + aiS , jS < 0.0
aiL, jS + aiS , jS = 0.0
0.0 < aiL, jS + aiS , jS ≤ 0.1
0.1 < aiL, jS + aiS , jS ≤ 0.2
0.2 < aiL, jS + aiS , jS ≤ 0.3
0.3 < aiL, jS + aiS , jS ≤ 0.4
0.4 < aiL, jS + aiS , jS ≤ 0.5
0.5 < aiL, jS + aiS , jS ≤ 0.6
0.6 < aiL, jS + aiS , jS ≤ 0.7
0.7 < aiL, jS + aiS , jS ≤ 0.8
0.8 < aiL, jS + aiS , jS ≤ 0.9
0.9 < aiL, jS + aiS , jS < 1.0
aiL, jS + aiS , jS = 1.0
0
31
149
8
3
2
3
0
0
0
0
0
0
1
33
144
10
2
4
1
1
0
0
0
0
0
1
29
150
7
5
3
1
0
0
0
0
0
0
0
17
164
6
6
2
1
0
0
0
0
0
0
0
16
167
6
5
2
0
0
0
0
0
0
0
2
126
774
37
21
13
6
1
0
0
0
0
0
Total
196
196
196
196
196
980
aiS , jS < 0
aiS , jS = 0
0 < aiS , jS ≤ 0.1
0.1 < aiS , jS ≤ 0.2
0.2 < aiS , jS ≤ 0.3
0.3 < aiS , jS ≤ 0.4
0.4 < aiS , jS ≤ 0.5
0.5 < aiS , jS ≤ 0.6
0.6 < aiS , jS ≤ 0.7
0.7 < aiS , jS ≤ 0.8
0.8 < aiS , jS ≤ 0.9
0.9 < aiS , jS < 1.0
aiS , jS = 1.0
0
49
140
5
2
0
0
0
0
0
0
0
0
1
49
138
6
2
0
0
0
0
0
0
0
0
1
30
156
6
3
0
0
0
0
0
0
0
0
0
19
169
6
2
0
0
0
0
0
0
0
0
0
20
168
7
1
0
0
0
0
0
0
0
0
2
167
771
30
10
0
0
0
0
0
0
0
0
Total
196
196
196
196
196
980
aiL, jS < 0
aiL, jS = 0
0 < aiL, jS ≤ 0.1
0.1 < aiL, jS ≤ 0.2
0.2 < aiL, jS ≤ 0.3
0.3 < aiL, jS ≤ 0.4
0.4 < aiL, jS ≤ 0.5
0.5 < aiL, jS ≤ 0.6
0.6 < aiL, jS ≤ 0.7
0.7 < aiL, jS ≤ 0.8
0.8 < aiL, jS ≤ 0.9
0.9 < aiL, jS < 1.0
aiL, jS = 1.0
0
36
153
3
2
1
1
0
0
0
0
0
0
1
38
150
4
1
1
1
0
0
0
0
0
0
0
32
156
6
1
0
1
0
0
0
0
0
0
0
19
169
6
1
1
0
0
0
0
0
0
0
0
18
172
5
0
1
0
0
0
0
0
0
0
1
143
800
24
5
4
3
0
0
0
0
0
0
Total
196
196
196
196
196
980
36
Table A2. (continued) Distribution of Input Coefficients by Year: Large Firms
(1)
1980
(2)
1985
(3)
1990
(4)
1995
(5)
2000
(6)
Total
aiL, jL + aiS , jL < 0.0
aiL, jL + aiS , jL = 0.0
0 < aiL, jL + aiS , jL ≤ 0.1
0.1 < aiL, jL + aiS , jL ≤ 0.2
0.2 < aiL, jL + aiS , jL ≤ 0.3
0.3 < aiL, jL + aiS , jL ≤ 0.4
0.4 < aiL, jL + aiS , jL ≤ 0.5
0.5 < aiL, jL + aiS , jL ≤ 0.6
0.6 < aiL, jL + aiS , jL ≤ 0.7
0.7 < aiL, jL + aiS , jL ≤ 0.8
0.8 < aiL, jL + aiS , jL ≤ 0.9
0.9 < aiL, jL + aiS , jL < 1.0
aiL, jL + aiS , jL = 1.0
0
29
150
8
5
3
0
1
0
0
0
0
0
0
28
153
5
5
2
2
1
0
0
0
0
0
0
27
154
7
3
2
2
1
0
0
0
0
0
0
14
168
6
4
2
2
0
0
0
0
0
0
0
15
167
7
3
2
2
0
0
0
0
0
0
0
113
792
33
20
11
8
3
0
0
0
0
0
Total
196
196
196
196
196
980
aiS , jL < 0.0
aiS , jL = 0.0
0 < aiS , jL ≤ 0.1
0.1 < aiS , jL ≤ 0.2
0.2 < aiS , jL ≤ 0.3
0.3 < aiS , jL ≤ 0.4
0.4 < aiS , jL ≤ 0.5
0.5 < aiS , jL ≤ 0.6
0.6 < aiS , jL ≤ 0.7
0.7 < aiS , jL ≤ 0.8
0.8 < aiS , jL ≤ 0.9
0.9 < aiS , jL < 1.0
aiS , jL = 1.0
0
49
140
7
0
0
0
0
0
0
0
0
0
0
45
141
10
0
0
0
0
0
0
0
0
0
0
29
157
10
0
0
0
0
0
0
0
0
0
0
18
169
9
0
0
0
0
0
0
0
0
0
0
19
169
8
0
0
0
0
0
0
0
0
0
0
160
776
44
0
0
0
0
0
0
0
0
0
Total
196
196
196
196
196
980
aiL, jL < 0.0
aiL, jL = 0.0
0 < aiL, jL ≤ 0.1
0.1 < aiL, jL ≤ 0.2
0.2 < aiL, jL ≤ 0.3
0.3 < aiL, jL ≤ 0.4
0.4 < aiL, jL ≤ 0.5
0.5 < aiL, jL ≤ 0.6
0.6 < aiL, jL ≤ 0.7
0.7 < aiL, jL ≤ 0.8
0.8 < aiL, jL ≤ 0.9
0.9 < aiL, jL < 1.0
aiL, jL = 1.0
1
34
149
8
3
0
0
1
0
0
0
0
0
0
33
154
5
2
1
0
1
0
0
0
0
0
0
31
157
5
1
1
1
0
0
0
0
0
0
0
18
170
5
1
1
1
0
0
0
0
0
0
0
19
170
4
1
1
1
0
0
0
0
0
0
1
135
800
27
8
4
3
2
0
0
0
0
0
Total
196
196
196
196
196
980
37
Table A3. Annual Data Used in Regression: Small firms
(1)
Cash / Sales
ratio
(2)
Debt / Sales
ratio
(3)
Borrowing
/ Sales ratio
(4)
Growth Rate
of Sales
(5)
Lending
Attitude
of banks
LIQ
DEBT 1
DEBT 2
S GROWT H
LEND
Food and beverages
1980
1985
1990
1995
2000
0.0870
0.0922
0.1228
0.1160
0.1125
0.3786
0.4090
0.5195
0.5720
0.4814
0.2239
0.2610
0.3556
0.4084
0.3417
0.1194
0.1128
-0.0289
0.2745
0.0961
-15.5
20.0
9.5
17.5
3.3
Textiles
1980
1985
1990
1995
2000
0.1117
0.1297
0.1366
0.2378
0.2256
0.5292
0.5473
0.6298
0.8965
0.7704
0.2731
0.2839
0.3521
0.5907
0.5396
-0.1088
0.1182
-0.1581
-0.1376
0.0179
-17.0
20.8
5.0
1.5
-17.0
1980
1985
1990
1995
2000
0.0781
0.1240
0.1168
0.1505
0.1279
0.3703
0.5069
0.5163
0.6955
0.7088
0.1141
0.2207
0.2599
0.4369
0.4459
-0.2269
0.2363
-0.2106
0.0523
-0.2704
1.8
33.3
17.8
20.8
10.3
1980
1985
1990
1995
2000
0.1062
0.1096
0.2039
0.1398
0.2837
0.4014
0.4748
0.5363
0.5292
0.5949
0.1449
0.1978
0.2714
0.2709
0.3780
-0.0507
0.0602
-0.1692
-0.3425
0.2420
-10.5
39.5
15.8
27.8
26.5
1980
1985
1990
1995
2000
0.0916
0.1186
0.1205
0.1142
0.1695
0.4063
0.4160
0.4563
0.4316
0.5063
0.1540
0.1851
0.2478
0.2252
0.2475
-0.3220
-0.2653
0.0259
-0.0316
-0.1544
-6.5
25.8
5.0
33.8
31.5
1980
1985
1990
1995
2000
0.1006
0.1285
0.1415
0.1665
0.1521
0.4525
0.6370
0.5855
0.8474
0.6980
0.1856
0.3404
0.2994
0.5343
0.4494
0.0609
-0.0815
-0.1180
-0.0796
0.1192
-11.3
18.3
10.3
13.8
-1.5
1980
1985
1990
1995
2000
0.1107
0.1592
0.1200
0.1326
0.1397
0.4663
0.5288
0.5490
0.7359
0.5512
0.2056
0.2224
0.3014
0.4527
0.3184
0.4174
0.1925
0.0620
0.1893
-0.3035
-13.3
23.0
12.5
19.3
-8.0
Pulp ,paper and
paper products
Chemicals
Petroleum and
coal products
Non-metallic
mineral products
Iron and steel
38
Table A3. (continued) Annual Data Used in Regression: Small firms
Non-ferrous
Metals
Fabricated metal
Products
Machinery
Electrical
machinery,
equipment and
Supplies
Transport
Equipment
Precision
Instruments
Miscellaneous
Manufacturing
(1)
Cash / Sales
ratio
(2)
Debt / Sales
ratio
(3)
Borrowing
/ Sales ratio
(4)
Growth Rate
of Sales
(5)
Lending
Attitude
of banks
LIQ
DEBT 1
DEBT 2
S GROWT H
LEND
1980
1985
1990
1995
2000
0.1018
0.0998
0.1178
0.1429
0.1248
0.3484
0.4440
0.4977
0.6239
0.5775
0.1448
0.1949
0.2614
0.3910
0.3368
-0.0981
0.0123
0.0830
0.4033
-0.4266
-16.8
33.0
21.8
23.3
-2.3
1980
1985
1990
1995
2000
0.1039
0.1243
0.1442
0.1435
0.1433
0.4403
0.4740
0.5343
0.6238
0.6766
0.1893
0.2333
0.2847
0.3744
0.4379
-0.0152
0.0610
-0.1141
0.1395
0.3582
-6.0
17.5
16.8
10.0
-0.5
1980
1985
1990
1995
2000
0.1373
0.1486
0.1344
0.1258
0.3575
0.5068
0.5484
0.5120
0.7249
0.7164
0.2118
0.2583
0.2564
0.4598
0.4577
0.0228
0.2339
-0.0894
-0.0859
0.0428
-9.8
19.8
18.0
8.8
-6.5
1980
1985
1990
1995
2000
0.0733
0.0924
0.0868
0.1191
0.1026
0.3465
0.3271
0.3915
0.5750
0.4856
0.1345
0.1390
0.2002
0.3414
0.2457
0.1834
0.0289
0.1623
0.0292
0.0604
-1.5
30.5
13.5
12.3
-1.5
1980
1985
1990
1995
2000
0.1013
0.1260
0.1171
0.1209
0.1362
0.4421
0.4843
0.4664
0.5125
0.6148
0.1947
0.2281
0.2272
0.2572
0.3900
0.3986
0.0935
-0.0038
0.0699
0.0779
-3.5
22.0
17.5
19.8
-1.3
1980
1985
1990
1995
2000
0.0984
0.1290
0.1745
0.1268
0.2070
0.3578
0.4778
0.5725
0.7689
0.5296
0.1518
0.2398
0.3255
0.5300
0.2862
0.0769
0.0715
0.0675
0.0842
0.3888
10.5
26.0
14.8
4.8
-3.8
1980
1985
1990
1995
2000
0.0960
0.1200
0.1212
0.1432
0.1738
0.4169
0.4758
0.5084
0.6396
0.5759
0.1825
0.2197
0.2760
0.3852
0.3485
-0.0534
0.2241
0.0016
-0.1014
0.1565
-12.5
19.7
13.7
8.4
-3.6
39
Table A3. (continued) Annual Data Used in Regression: Large firms
(1)
Cash / Sales
ratio
(2)
Debt / Sales
ratio
(3)
Borrowing
/ Sales ratio
(4)
Growth Rate
of Sales
(5)
Lending
Attitude
of banks
LIQ
DEBT 1
DEBT 2
S GROWT H
LEND
Food and beverages
1980
1985
1990
1995
2000
0.0935
0.1208
0.1668
0.1484
0.1100
0.3946
0.3908
0.4094
0.4369
0.4579
0.1452
0.1155
0.1120
0.1374
0.1654
0.0101
0.2072
0.0514
0.0172
0.0390
-14.3
46.2
8.6
33.8
18.4
Textiles
1980
1985
1990
1995
2000
0.1474
0.1403
0.1511
0.1567
0.1933
0.6750
0.6555
0.8717
0.8833
1.1075
0.3385
0.3306
0.4223
0.4171
0.6309
0.0322
0.0401
0.1771
0.0149
-0.0288
-23.9
42.1
-3.3
23.0
2.0
1980
1985
1990
1995
2000
0.1315
0.1410
0.1485
0.1006
0.0909
0.7454
0.8036
0.8454
0.9218
0.8848
0.3710
0.4246
0.3662
0.4663
0.4609
-0.0947
-0.0282
0.0390
0.0653
0.0401
-28.4
35.5
-12.8
24.0
24.2
1980
1985
1990
1995
2000
0.1304
0.1497
0.2259
0.2183
0.2143
0.6274
0.6450
0.6848
0.7298
0.6721
0.2526
0.2607
0.1940
0.2662
0.2236
-0.0680
0.0386
0.0780
0.0198
0.0173
-21.1
43.7
0.3
31.4
25.1
1980
1985
1990
1995
2000
0.0662
0.0634
0.1000
0.0951
0.0486
0.5544
0.5883
0.5786
0.6775
0.5522
0.3322
0.3502
0.3353
0.3674
0.2813
-0.1624
0.1450
0.0643
-0.0203
0.0134
-38.1
29.0
-27.2
25.5
16.2
1980
1985
1990
1995
2000
0.1665
0.1909
0.1974
0.1706
0.1661
0.6860
0.7344
0.6571
0.7470
0.8666
0.3297
0.3608
0.2173
0.3061
0.3639
0.0017
-0.0036
-0.0381
-0.0537
0.0297
-22.3
30.1
-3.9
30.1
11.7
1980
1985
1990
1995
2000
0.1391
0.1721
0.1820
0.1805
0.1260
0.9181
1.0922
0.8499
1.1203
1.1380
0.4247
0.5165
0.2730
0.4447
0.5056
-0.0044
-0.0263
0.0503
-0.0090
0.0163
-26.3
40.9
-6.2
27.8
1.7
Pulp ,paper and
paper products
Chemicals
Petroleum and
coal products
Non-metallic
mineral products
Iron and steel
40
Table A3. (continued) Annual Data Used in Regression: Large firms
Non-ferrous
Metals
Fabricated metal
Products
Machinery
Electrical
machinery,
equipment and
Supplies
Transport
Equipment
Precision
Instruments
Miscellaneous
Manufacturing
(1)
Cash / Sales
ratio
(2)
Debt / Sales
ratio
(3)
Borrowing
/ Sales ratio
(4)
Growth Rate
of Sales
(5)
Lending
Attitude
of banks
LIQ
DEBT 1
DEBT 2
S GROWT H
LEND
1980
1985
1990
1995
2000
0.1179
0.1418
0.1250
0.1147
0.0918
0.6936
0.7261
0.6077
0.8492
0.9431
0.3689
0.4039
0.2198
0.4530
0.4819
0.0149
-0.0297
0.0200
0.0538
0.0587
-35.9
38.0
-6.1
37.8
12.0
1980
1985
1990
1995
2000
0.1316
0.1413
0.1751
0.1783
0.1679
0.6155
0.6005
0.5925
0.6652
0.6850
0.2487
0.2429
0.1928
0.2467
0.2576
0.0294
-0.0193
0.0080
0.0612
-0.0116
-2.5
37.6
11.3
32.8
11.9
1980
1985
1990
1995
2000
0.1741
0.2091
0.2167
0.2235
0.1926
0.6613
0.6659
0.6209
0.7088
0.7060
0.2280
0.2092
0.1584
0.2188
0.2370
0.1224
0.0599
0.0788
0.0357
0.1126
-15.4
42.4
7.1
25.5
5.9
1980
1985
1990
1995
2000
0.1242
0.1487
0.2030
0.1750
0.1257
0.4881
0.4972
0.5091
0.5308
0.4985
0.1202
0.0898
0.0901
0.1162
0.0952
0.1728
0.0669
0.1610
0.1462
0.1698
-7.2
40.8
10.0
22.4
19.2
1980
1985
1990
1995
2000
0.1224
0.1146
0.1365
0.1306
0.1250
0.5552
0.4757
0.4471
0.5079
0.5265
0.1927
0.1501
0.0953
0.1259
0.1216
0.1194
0.0848
0.1033
0.0216
0.0678
-10.7
45.6
4.7
34.0
9.0
1980
1985
1990
1995
2000
0.1345
0.1823
0.2361
0.2097
0.1441
0.4806
0.4840
0.5698
0.5833
0.5176
0.1334
0.1354
0.1487
0.1646
0.1652
0.1855
0.1371
0.0871
0.0542
0.1090
4.0
45.8
3.4
26.3
19.8
1980
1985
1990
1995
2000
0.1201
0.1420
0.1760
0.1514
0.1728
0.4764
0.5035
0.5390
0.5616
0.5733
0.1568
0.1573
0.1548
0.1973
0.2007
-0.0273
0.1217
-0.0018
0.0594
-0.0463
-8.7
34.5
7.1
26.4
12.0
41